RisCO2: Implementation and Performance Evaluation of RISC-V Processors for Low-Power CO₂ Concentration Sensing.

Area of Science:

• Embedded Systems Engineering
• Computer Architecture
• Digital Signal Processing

Background:

• Energy efficiency is paramount for battery-powered embedded systems.
• RISC-V processors offer flexibility but vary in energy performance.
• Evaluating processors for specific applications like CO2 sensing is crucial.

Purpose of the Study:

• To introduce RisCO2, an RISC-V processor optimized for energy efficiency.
• To evaluate RisCO2's performance against other RISC-V cores in NDIR CO2 sensor signal processing.
• To assess energy consumption and resource utilization of different RISC-V implementations.

Main Methods:

• Implementation of RisCO2 and four other RISC-V processors (Ri5cy, Micro-riscy, Zero-riscy, CV32E40P) on the PULPino SoC.
• Synthesis using Vivado IDE for performance evaluation.
• Testing in a signal processing application for nondispersive infrared (NDIR) CO2 sensors.

Main Results:

• RisCO2 demonstrated the lowest energy consumption, with reductions of 53.5% vs. CV32E40P and 94.8% vs. Micro-riscy.
• RisCO2 achieved the lowest FPGA resource utilization, showing a 46.1% reduction in LUTs compared to CV32E40P and 59% vs. Ri5cy.
• The RV32E_Zfinx based RisCO2 was specifically designed for low-power signal demodulation.

Conclusions:

• RisCO2 is a highly energy-efficient RISC-V processor suitable for NDIR CO2 sensors requiring accurate signal demodulation.
• Processor evaluation in specific use cases is vital for identifying optimal energy-efficient solutions.
• Findings provide valuable insights for designing energy-efficient embedded systems with RISC-V.